anti-jamming performance of hybrid fec code in the ... performance of hybrid fec code in the...

International Journal of Novel Research in Engineering and

Applied Sciences (IJNREAS) 1(1) February 2014

©Avi-D Publishers 2014

www.avidpublishers.ca

Anti-jamming Performance of Hybrid FEC code in the Presence of CRN

Random Jammers

Victor Balogun Computer Science Department

University of Idaho

Moscow, ID, USA

[email protected]

Abstract

The survivability of Cognitive Radio Networks (CRNs) operating in the presence of malicious attackers,

especially jammers is a crucial security issue. Cognitive Radio (CR) jammers are capable of taking

advantage of the reconfigurable features of CRN so as to cause faults of different severity including value

faults. The performance of CRNs under jamming attacks using the hybrid fault-model approach which

considers different fault scenarios have been investigated. A hybrid Forward Error Correction (HFEC)

code was proposed so as to mitigate the observed high jamming impact of CR jammers. The proposed

HFEC code was defined as a concatenation of the Raptor code and the Secure Hash Algorithm-2 (SHA-2).

The Raptor part is used to recover data loss due to jamming, while SHA-2 is used to verify the integrity of

data received at the destination since CR jammers are capable of manipulating transmitted data leading to

value faults. The HFEC code was designed in this manner so as to make it capable of handling all

jamming scenarios identified under the hybrid fault model classification. In a previous paper, the

performance of the HFEC code in a CRN operating in the presence of CR constant jammers was

quantified. It was found that the HFEC code was robust against all instances of constant jammers,

especially scenarios where value faults are introduced as a result of manipulated data. The need to

investigate the performance of the HFEC in the presence of other jamming types like random jammers is

imperative. This is because the random jammers behave significantly different from the constant jammers

in their mode of operation. In this paper, we present the performance of the HFEC code in NS-2 extended

for CRN. We essentially evaluate and analyze its performance against CR random jammers which are

different from the constant jammers using relevant performance metrics. The observed high Packet

Delivery Ratio (PDR) and recovery rate of the algorithm for all simulated scenarios show that the

encoding and decoding algorithm of the proposed HFEC code is very efficient and resilient against the

different rate of jamming of random jammers.

mailto:[email protected]

2

International Journal of Novel Research in Engineering and Applied Sciences (IJNREAS) 1(1) February 2014

Keywords: Jamming, Forward Error Correction Codes, Cognitive Radio Networks, Fault models.

1.0 INTRODUCTION

Cognitive Radio (CR) (Akyildiz, et al., 2006)

concept which allows incumbent

(primary/licensed) users to share their spectrum

with unlicensed (secondary) users in a non-

interference manner has been regarded as a way

out of the problem of spectrum

shortage/underutilization (Mitola III, 2000). This

problem emanates from the fact that most of the

frequency spectrum has been allocated to licensed

users, even though these users do not necessarily

occupy these frequencies all the time. An

example of this is television stations. The free

spectrum that are not licensed are congested

because most of the mobile devices operate at

these free frequencies. With the CR technology,

the unlicensed users can opportunistically make

use of the licensed spectrum during the OFF

period of the incumbents but they must vacate it

immediately the incumbents are present.

Accurate sensing of spectrum for the presence of

an incumbent is a very crucial issue in CR

networks as detection could be significantly

affected by impairments such as shadowing and

multipath fading. Cooperative Spectrum Sensing

(CSS) (Akyildiz, et al., 2011) has been proposed

and found to be effective in mitigating these

impairments. In spite of the anticipated success

and potentials of CSS, the presence of malicious

users like jammers operating in the

neighbourhood of a CSS Cognitive Radio

Network is of great security concern. At the top

level of this concern is that CRs have flexible and

adaptable features that make them easy prey to

security threats like jamming attacks, Primary

User Emulation (PUE) attacks, masquerading of

CR, Spectrum Sensing Data Falsification (SSDF)

attacks and many more that are either specific to

CRs or inherited from traditional wireless

networks.

Jamming, which is the main focus of this paper,

has been addressed by different authors (Dong &

Liu, 2010, Pelechrinis et al., 2011, Tague, 2010,

Xu et al., 2005) with respect to traditional

wireless networks and different anti-jamming

techniques have been proposed. There is a need

to investigate the performance of CRs in the

presence of jammers since CR networks differ

significantly from traditional wireless networks.

Though several authors (Asterjadhi, 2010,

Burbank et al, 2008, Cadeau & Li, 2012, Yongle

et al., 2012) have published work on jamming in

CR Networks, to the best of our knowledge none

have studied jamming in the context of fault

model classifications (including value faults) and

their respective fault handling.

3


The CRN was classified as a network that could

experience different mixes of faults especially

one that considers value faults using the hybrid

fault model classification in (Balogun & Krings,

2013). The perceived impacts of jamming on this

model was quantified in (Balogun & Krings,

2014a), and in (Balogun & Krings, 2014b) we

proposed and presented the algorithmic design of

a new hybrid forward error correction (FEC) code

that is capable of mitigating jamming attacks

under this classification.

We also presented the implementation of the

proposed solution in Network Simulator 2 (NS-2)

and the analysis of its performance against CR

constant jammers. In this paper, we present the

performance evaluation of the proposed hybrid

FEC code in the presence of a different class of

jammer i.e. CR random jammer. This is because

the behavior of random jammers is different from

that of the constant jammers.

This will help to verify the suitability of the

proposed anti-jamming solution in mitigating CR

random jammers which have the potential of

alternating between sleeping and jamming in

order to maximize their jamming effectiveness.

The observed robustness of the hybrid FEC code

with its effectiveness and computational

feasibility presents a practical solution to the

survivability of CRNs under any jamming

attacks.

2.0 COGNITIVE RADIO RANDOM

JAMMERS

Jammers in traditional wireless communications

can cause Denial of Service (DoS) at either the

transmitter or the receiver if they adequately

inject interfering signals into the same spectral

region as that of the legitimate users (Dong &

Liu, 2010). A random jammer is a class of

jammer that alternates between jamming and

sleeping instead of jamming continuously like a

constant or deceptive jammer. It has the potential

to behave either as a constant jammer

(continuously sending random signals) or a

deceptive jammer (continuously sending regular

packets) during the wake period. The sleep and

wake period which could be either fixed or

random values, can be used as threshold values

between power consumption and jamming

efficiency.

The random jammer is also different from a

reactive jammer which transmits signals only

when it finds the channel to be busy so as to

cause DoS to an on-going transmission.

A Cognitive Radio operating as a random jammer

is able to cause DoS to all the CSS Architectural

types discussed in (Akyildiz, et al., 2011) and

(Balogun & Krings, 2013) depending on its

location and power. Figure 1 illustrates jamming

in a combined CSS Cognitive Radio Network

4


(Balogun & Krings, 2013). The impact of the

jamming node is shown in the figure by the

communication it impacts. As depicted in the

figure, the jamming malicious node was able to

transmit jamming signals to all the CRs operating

within the reach of its signal depending on the

signal power. The jamming node also jams

signals from the primary user base station

because it has no regard to obey the non-

interference to primary user policy of the CRNs

standard.

Impact of Jamming attack in CRNs:

The impact of different classes of jammers

operating in a CRN environment has been

quantified in (Balogun & Krings, 2014a). Table I

presents a summary of the comparison of these

jammers i.e. Constant (Con), Random (Ran),

Deceptive (Dec) and Reactive (Rec) with a rating

of their impact. The table is intended to give the

reader a feeling for the impact and what to expect

when trying to mitigate specific jamming attacks

as found in cited literature and the results of the

simulation we described in (Balogun & Krings,

2014a).

5


The simple values low, medium, high or

adjustable are used in the network metrics. The

metrics considered are:

(i) The power usage of the jammer during the

entire jamming duration. (ii) The throughput,

which is the total number of packets delivered to

the receiver during the simulation period by the

CRN experiencing jamming. (iii) The cost of

attack, which is the total cost of perpetrating the

jamming attack. This involves generally

cost/complexity of designing the jammer. (iv)The

cost of defense, which is the cost/complexity of

designing anti-jamming techniques to handle the

jamming. (v) Scalability, which refers to the

impact of increasing number of the jammers. (vi)

Level of DoS, which is the level of disruption

caused by the jammers in the CRN. (vii)

Detection probability, which is the probability of

the jammers being detected by the CRN as they

carry out jamming.

3.0 DEALING WITH CR RANDOM

JAMMERS – THE HYBRID FEC CODE

A review (Balogun & Krings, 2014b) of several

anti-jamming strategies that have been proposed

over the years shows that the solution strategies

proposed by the authors are only suitable to deal

with benign and omissive faults in a fault-model

classified CRN. Since the CR jammers are also

capable of introducing transmissive value faults

in pathological cases we therefore proposed the

hybrid FEC code. The hybrid FEC code is

defined by the concatenation of Raptor code

(Shokrollahi, 2006) and SHA-2 (Sklavos &

Koufopavlou, 2005). The Raptor part of the code

is used to recover data loss due to omissive fault,

while the SHA-2 hash function is used to detect

transmissive (value) faults.

The Raptor code is an FEC code with linear time

encoding and decoding that allows a message

made up of a number of k symbols to be encoded

into an infinite series of symbols in such a way

that if during transmission some part of the data

6


is lost, e.g., due to jamming, the lost data can be

recovered with a non-zero probability that

increases as the number of the received symbols

increases beyond k (Shokrollahi, 2006). The

Raptor code is made up of two essential

components: (i) Raptor encoder - which is used to

encode the message at the sender side (ii) Raptor

decoder - which is used to recover back the

original message at the receiver side. SHA-2 is a

computationally efficient hash function that can

be used to verify the integrity of data by

transforming any set of data elements into a

unique fixed length hash value known as the

message digest. For instance a message M of

length l can be hashed with SHA-256 (a type of

SHA-2) to produce a 256-bit message digest

provided 0 ≤ l < 264

.

The advantages of using SHA-2 are: (i) SHA-2 is

collision resistant, meaning that the probability of

finding two messages (no matter the similarities)

that will hash to the same hash value will require

work equivalent to 2n/2

hash computations

(Bakhtiari et al., 1995, Sklavos & Koufopavlou,

2005). (ii) SHA-2 computation is a one-way

operation – the one-way operation characteristic

which is the basis of SHA-2 security implies that

for a given hash value, it should require 2n hash

computation in order to find any message that

hashes to that same value (Bakhtiari et al., 1995).

(iii) A hash value can be distributed or stored

without any need for encryption since it is used

for comparative purpose only.

3.1 Mode of Operation

The hybrid FEC code operates between the

application layer and the transport layer. Before a

sender’s message is sent from the application

layer to the transport layer, the SHA-2 module is

used to generate the message digest. This

message digest is inserted into the message’s

packet header and the message is passed to the

encoder part of the Raptor module.

At the pre-code stage of the encoder, redundant

symbols are added to the message while at the LT

code (Luby, 2002) stage, the LT code is used to

generate the encoded output symbols. These

encoded output symbols are passed down to

lower layers of the protocol stack. At the receiver,

when the message arrives at the transport layer,

the Raptor decoder starts the decoding process as

soon as it receives about k(1 + ε) of the encoded

output symbols. k(1 + ε) is the minimum amount

of the encoded symbols that should be received

by the receiver before it can start the decoding

process, where " is the decoding inefficiency or

overhead of the Raptor code. Once the decoding

process is successful, the decoded message is sent

to the SHA-2 module where the message digest

of the decoded message is generated. The

message digest generated is then compared with

7


the message digest stored in the packet’s header

of the original message. If the two digests match,

then the message is received with no error and

forwarded to the receiver through the application

layer. If there is a difference between these two

message digests, then it means that the message

has been manipulated along the transmission

path.

3.2 Jamming Recovery Process

When communicated data is detected to be lost or

manipulated, the approach explores a recovery

block (Balogun & Krings, 2014a) based on the

application scenario. The recovery block provides

an efficient recovery means and safety net for the

hybrid code when jamming, especially one that

produces value fault is detected. The jamming

recovery process is as follows:

If the decoding fails due to loss of encoded

data, the decoder waits for more encoded

symbols to arrive and tries again the decoding

process.

If no more encoded symbol is received,

jamming is suspected.

If the decoding is successful, the decoded

message is passed to the message digest

generator and the hash value is compared

with that of the sent message.

Jamming is suspected if: (i) there is a

decoding failure of the Raptor code (ii) the

message digests do not match.

The recovery block is explored to recover

from the attack.

We investigated the performance of this new

solution against constant jammers in Fault-

Model-Classified CRN and found that the

solution was robust against all the jamming

scenarios identified in the model (Balogun &

Krings, 2014b). In this paper, however, we are

only interested in investigating its performance

against the CR random jammers whose behavior

is different from the constant jammers. For

details about the hybrid FEC code algorithm and

implementation, the reader is referred to

(Balogun & Krings, 2014b).

4.0 SIMULATION

4.1 Assumptions

Some assumptions made for the experiment

conducted include (i) Each jamming rate is

derived as a cumulative effect of jamming

activities from one or more jammers. This means

that the jamming rate is a threshold function

whose impact do not increase beyond the

threshold even if more than one jammer is present

in the network. (ii) Fusion center or base station

for the CRN is trusted and therefore cannot

become a jamming CR node.

4.2 Simulation Parameters

Table II presents the different simulation

parameters that were used in simulating the

8


hybrid FEC code. Other parameters used that were not specific to NS-2 include:

Number of input symbols k in a block: Based

on the analysis in (Eittenberger & Mladenov,

2012), we use block size k = 1000symbols for

our simulation, representing a not too large or

too small block size.

Length of input symbol l in bytes: The Raptor

code has been defined as a code with a linear

decoding time of O(k log(1/ε)), it theoretically

follows that this decoding time of the Raptor

code could be decreased significantly if a large

symbol size l and a large block size is chosen.

Based on the analysis in (Eittenberger &

Mladenov, 2012), we use a length of 1450

bytes which is the maximum UDP packet size

we specified for our simulation.

Number of Repair symbol ε: The number of

repair symbols that minimizes the overhead for

our chosen block size k = 1000 was discovered

to be about 16 repair symbols. The experiment

in (Eittenberger & Mladenov, 2012) shows

that 16 repair symbols used for a block size of

about 1000 symbols reduces the overhead to

about 1.56%. We therefore use 16 as the

number of repair symbols for our simulation.

5.0 ANALYSIS OF SIMULATION RESULTS

FOR RANDOM JAMMERS

Random jamming scenarios are simulated in a

Combined CSS CRN with both the ordinary FEC

code and the hybrid FEC code. Combined CSS

architectural model that considers both omissive

and transmissive value faults is deployed. The

parameters of measurement are (i) Throughput -

defined as the average rate of data successfully

delivered by the network (ii) Packet Delivery

9


Ratio (PDR) - defined as the ratio of the sum of

packets received at a layer to the sum of packets

sent in the same layer and (iii) Packet Loss Ratio

(PLR) - defined as Number of lost packet /

(Number of lost packet + Number of packets

received successfully). The result of our analysis

is as follows:

5.1 Throughput

Figures 2 and 3 depict the random jamming

scenario of a CRN experiencing a jamming rate

of 10%. It is noticeable from Figure 3 that the

hybrid FEC code was able to mitigate every

jamming effect caused by the random jammer,

but the ordinary FEC code in Figure 2 fails in

some intervals characterized by value faults

caused by the random jammers. The throughputs

of both the ordinary FEC and the hybrid FEC

codes were very high because the random jammer

alternates between sleeping and jamming, and as

such the impact of the jamming was contained by

both ordinary and hybrid FEC code.

10


Figures 4 and 5 depict the random jamming

scenario when the rate of jamming was 50%. It is

noticeable from the figure that the hybrid FEC

code was able to mitigate most jamming effect

caused by the random jammer but fails in some

intervals due to the actions of the recovery block.

In the case of ordinary FEC code in Figures 4, the

throughput of the network dropped significantly

as the code fails at every instance of transmissive

value fault and whenever there is a decoding

failure of the FEC code. This trend is noticed at

jamming rate between 50% - 90%. Beyond 90%

jamming rate, the throughput of ordinary FEC

drops to zero while that of hybrid FEC remains

moderate even when the rate of jamming is

100%, as seen in Figure 6.

These figures show that the hybrid FEC code out-

performs the ordinary FEC code as moderately

high throughput was realized by the hybrid FEC

code while ordinary FEC code’s throughput

drops to zero when a CRN is subjected to a

jamming rate of 100%.

11


12


5.2 Packet Delivery Ratio

Figure 7 shows the PDR of a CRN implementing

ordinary FEC compared with that of the hybrid

FEC code. It is noticeable that the PDR of the

ordinary FEC code continues to drop gradually as

the rate of jamming increases because of the

sleeping and jamming schedule of the random

jammer. At 90% jamming, the PDR of the

ordinary FEC is about 15%. Beyond the jamming

rate of 90%, the PDR of ordinary FEC drops to

zero. The hybrid FEC maintains very high PDR

of more than 70% even when the jamming rate

increases to 100%.

13


5.3 Packet Loss Ratio

Figure 8 shows average packet loss ratio (PLR)

of a CRN under different random jamming attack

rates. It is observable from the figure that the

hybrid FEC code consistently maintains very low

PLR for all the jamming rate of the random

jammers. Similarly, the ordinary FEC code has

low PLR up to a jamming rate of about 50%, but

beyond this point the code breaks down

significantly as the PLR increases to about to

65%, even when the jamming rate is just about

60%. This emonstrates that the hybrid FEC code

maintains high recovery rates even when the rate

of jamming is as high as 100%, showing that the

algorithms of the hybrid FEC code are

computationally efficient and robust against any

level of random jamming and out-performs the

ordinary FEC code.

5.4 Summary of Statistical Analysis for

Random Jamming in a Combined CSS CRN

Tables III to V present the summary of the

statistical analysis of the result of the hybrid FEC

code (HFEC) simulation under random jamming

in a CR network. We only present the 10%

jamming representing the fair case of random

jamming, the 50% jamming representing the

moderate case of random jamming, and the 100%

representing the worst case of random jamming.

14


15


6.0 CONCLUSIONS

In this paper, we investigated the performance of

the proposed hybrid FEC code in a fault-model

classified CRN in the presence of CR random

jammers that are capable of introducing value

faults. The hybrid FEC code was found to be

superior in performance compared to the ordinary

FEC code because it provides higher throughput

and PDR. The proposed solution was robust

against all instances of random jammers even at

their worst case scenario of 100% jamming rate.

The algorithm of the hybrid code was also found

to be very efficient as it maintains high recovery

rates by providing consistent low packet loss ratio

16


(PLR), even at 100% jamming rate. The recovery

block provides an efficient recovery means and

safety net for the hybrid code when jamming that

produces value fault is detected.

REFERENCES

1. Akyildiz, I. F. et al. (2006). NeXt

generation/ dynamic spectrum access

/cognitive radio wireless networks: A

survey, Computer Networks 50 (13) 2127-

2159.

2. Akyildiz, I. F., Lo, B. F. & Balakrishnan,

R. (2011). Cooperative Spectrum Sensing

in Cognitive Radio Networks: A Survey,

Physical Communication (Elsevier)

Journal, vol. 4, no. 1, pp. 40-62.

3. Asterjadhi, M. Zorzi. (2010). JENNA: A

Jamming Evasive Network coding

Neighbor discovery Algorithm for

Cognitive Radio Networks, In IEEE ICC

Workshop on Cooperative and Cognitive

Mobile Networks (CoCoNet3), Cape

Town, South Africa.

4. Burbank, J. L. et al. (2008). A common

lexicon and design issues surrounding

cognitive radio networks operating in the

presence of jamming, IEEE MILCOM; 1-

7.

5. Balogun V. & Krings A. (2013). On The

Impact of Jamming Attacks on

Cooperative Spectrum Sensing in

Cognitive Radio Networks, Proc. 8th

Annual Cyber Security and Information

Intelligence Research Workshop, Oak

Ridge National Laboratory, TN, U.S.A.

January 8 - 10, 2013.

6. Balogun V. & Krings A. (2014a). An

Empirical Measurement of Jamming

Attacks in CSS Cognitive Radio

Networks, Accepted for presentation at

the 27th IEEE Annual Canadian

Conference on Electrical and Computer

Engineering (CCECE 2014), Toronto,

Canada, May 5 - 8, 2014.

7. Balogun V. & Krings A. (2014b).

Mitigating Constant Jamming in

Cognitive Radio Networks using Hybrid

FEC Code, Accepted for presentation at

the 28th IEEE International Conference

on Advanced Information Networking and

Applications (AINA-2014) Victoria,

Canada, May 13 - 16, 2014.

8. Cadeau, W. and Li, X. (2012). Anti-

jamming performance of cognitive radio

networks under multiple uncoordinated

jammers in fading environment, Proc. of

the 46th Annual CISS, Princeton Univ.,

NJ, March 2012.

9. Dong Qi & Liu Donggang. (2010).

Adaptive Jamming-Resistant Broadcast

Systems with Partial Channel Sharing,

Proc. 2010 International Conference on

17


Distributed Computing Systems, Genova,

Italy.

10. Eittenberger, P. M. & Mladenov, T.

(2012). Raptor Codes for P2P Streaming,

20th Euromicro International Conference

on Parallel, Distributed and Network-

Based Processing (PDP), 2012

11. Mitola III, J. (2000) Cognitive Radio: An

Integrated Agent Architecture for

Software Defined Radio, Doctor of

Technology dissertation, Royal Inst.

Technol. (KTH), Stockholm, Sweden,

2000.

12. Pelechrinis, K., M. Iliofotou and S. V.

Krishnamurthy, Denial of Service Attacks

in Wireless Networks: The case of

Jammers, In IEEE Communication

Surveys and Tutorials, pp(99):113, April

2011.

13. Shokrollahi, A. (2006). Raptor codes,

IEEE/ACM Transaction on Networking,

14(SI):2551-2567.

14. Bakhtiari, S.; Safavi-Naini, R.; and

Pieprzyk, J. Cryptographic Hash

Functions: A Survey. Technical Report

95-09, Department of Computer Science,

University of Wollongong, July 1995.

Available:

http://citeseer.ist.psu.edu/bakhtiari95crypt

ographic.html

15. Sklavos, N. & O. Koufopavlou. (2005)

Implementation of the SHA-2 Hash

Family Standard Using FPGAs, J. of

Supercomputing, Vol. 31, No 3, pp. 227-

248, 2005.

16. Luby, M. (2002) LT - Codes, in

Proceedings of the 43rd Annual IEEE

Symposium on the Foundations of

Computer Science (FOCS), pp. 271-282.

17. Tague, P. (2010). Improving anti-

jamming capability and increasing

jamming impact with mobility control, In

6th IEEE International Workshop on

Wireless and Sensor Networks Security

(WSNS), Nov. 2010.

18. Yongle Wu, Beibei Wang, K. J. Ray Liu,

and T. Charles Clancy, Anti-Jamming

Games in Multi-Channel Cognitive Radio

Networks. IEEE Journal on Selected

Areas in Communications, Vol. 30, NO.

1, January 2012

19. Xu, Wenyuan et al. The feasibility of

launching and detecting jamming attacks

in wireless networks, Proceedings of the

6th ACM international symposium on

Mobile ad hoc networking and computing,

May 25-27, 2005, Urbana-Champaign, IL,

USA.

http://citeseer.ist.psu.edu/bakhtiari95cryptographic.html

http://citeseer.ist.psu.edu/bakhtiari95cryptographic.html

anti-jamming performance of hybrid fec code in the ... performance of hybrid fec code in the...

Documents